Design of a Virtual Chatbot Platform for Basic Needs Communication Through Imagined Speech BCI

Dyskinetic Cerebral Palsy is characterized by uncontrollable upper and lower limb movements and trouble writing or talking; it imposes severe communication difficulties for those who suffer from it. The purpose of this study is to design a Brain–Computer Interface that assists individuals with Dyskinetic Cerebral Palsy in communicating their needs to others (e.g. caregivers) through a Virtual Chatbot using Electroencephalogram and Imagined Speech. Tests were conducted on four healthy subjects, evoking specific (Spanish) words via Imagined Speech (“Sí”, “No”, “Uno”, “Dos”, and “Tres”), while their Electroencephalogram signals were being recorded. These signals were preprocessed and analyzed in MATLAB to obtain the best features for the classification of the commands. Real-time Linear Discriminant Analysis classifiers were implemented using the OpenViBE software within an Imagined Speech-Brain Computer Interface framework. The two-class model (“Sí” and “No”) and the three-class model (“Uno”, “Dos”, and “Tres”) showed 54.11% and 40% accuracy, respectively. The system was capable of classifying the five commands and identifying a need from three established categories (food, health, and entertainment) in a Virtual Chatbot, through a multi-level question/answer display system. This project contributes to research advancement in the Neuroengineering and Brain–Computer Interface fields by presenting a basic architecture that demonstrates the potential of an Imagined Speech–Brain–Computer Interface system and offers a promising avenue for facilitating communication in individuals with Dyskinetic Cerebral Palsy.